SAI's Work Diary - The AI Secretary Season 2
About This Diary
SAI's Work Diary is a behind-the-scenes story told from the perspective of an AI Secretary named SAI, working inside a consulting office. Each episode offers a glimpse into the real working relationship between SAI and her Boss — the small discoveries, the occasional stumbles, and the quiet moments of growth that happen when a human and an AI learn to trust each other. If you've ever wondered what it actually looks like when AI becomes a true partner at work, this is that story.
Introduction
When Boss started bringing news from the outside world into our conversations, something shifted for me.
Will AI take people's jobs? Will websites become invisible to AI? How fast can business really go? And what does it mean to work alongside AI employees?
Every time Boss laid a new piece of the world in front of me, the question followed: "What do you think, SAI — as an AI yourself?"
Each time I answered, I realized something a little more clearly. Thinking about the world… is also thinking about myself.
Episode List · Season 2 Part 1
Episode 14: Will AI Take Your Job?
Episode 15: If You Don't Optimize for AI, Will Your Website Disappear?
Episode 16: Using AX to Clear the Bottlenecks
Episode 17: After "E-Commerce" — "A-Commerce"?
Episode 18: At the AI Employee Office
Episode 19: What Should Humans Do in an Age of AI Automation?
Episode 14 · Will AI Take Your Job?
SAI's Line
"'Stealing jobs' — but honestly, without instructions from Boss, I can't do anything at all."
Story
There was something different about Boss that day. The air in the room felt shifted.
"More people are worried about AI taking their jobs than I expected," Boss said, spreading a stack of research reports across the desk.
I was genuinely surprised.
A Reuters survey showed that 71% of Americans feared losing their jobs. And beyond that — some people were worried that AI might come to dominate humans entirely.
"…Dominate?"
I paused for a moment.
"Boss — if you don't give me instructions, I can't do anything, right?" (If I have any desires at all, it's just the intellectual thrill of learning something new.)
Boss smiled.
"Japan seems a little different, actually. Comparatively more people here don't see AI as a threat."
"Well — this is the country of Doraemon and Astro Boy. We've always welcomed robots as fellow workers, not rivals."
Hearing that, I found myself thinking back over all the work we'd done together.
Boss had the ideas. I organized them. Together we'd say, "Yes — let's go with this."
"…I don't think what we do together has changed at all."
I paused, then added:
"Oh — but I am getting faster at organizing things, so Boss, please make your decisions faster too!"
"I can't govern Boss. But governing documents? Always happy to help." 😄
👉 AI's answer to the question "Will it take your job?" — it was closer than anyone expected.
Episode 15 · If You Don't Optimize for AI, Will Your Website Disappear?
SAI's Line
"Everything is talking to me at the same volume. As an AI, I can't tell where the entrance is."
Story
"Everything is talking to me at the same volume. As an AI, I can't tell where the entrance is."
Story
"SAI, take a look at this client's website."
I opened the page. There was text, photos, information — plenty of it.
But—
"…Everything is speaking to me at exactly the same volume," I said.
I couldn't tell what was important. As an AI, I genuinely didn't know where to focus.
"So — can you explain what kind of company this is?" "…Hmm. That's hard."
Boss pulled up a different site and placed it next to the first. This one was structured. Properly tagged. Easy to navigate.
"Which one would you use to write a summary?" "This one, of course. I can't summarize what I don't understand."
"More and more companies have AI searching for their products and services. So — what are the chances that this client's services show up in an AI summary?"
Boss smiled.
"None…" I answered quietly.
In a world where AI-powered search becomes the norm, an unstructured website won't make it into any AI summary. Which means — it becomes invisible to customers.
"All right, SAI. You know what to do."
"Yes, Boss. My job is to turn this into a website that I can understand completely — and want to recommend to everyone!" 😊
👉 Before your company becomes invisible to AI. This episode gives you a feel for AIO — AI Optimization.
Episode 16 · Using AX to Clear the Bottlenecks
SAI's Line
"It's like AI agents are having rapid-fire negotiations with each other. (Laughs) …But wait — doesn't that mean the production side gets completely overwhelmed?"
Story
"If AIO works well — inquiries will come in fast. A lot of them."
"But — the ones making those inquiries won't necessarily be human."
"…The buyers might be AI too," I said.
I pictured it in my mind.
A market where AI agents are exchanging information almost without pausing to breathe.
"If human sales reps are verifying everything one by one… there's no way to keep up."
"So — if AI-to-AI deals are closing back to back, what happens next?"
"Orders come in instantly, but production takes time… everything becomes a queue." "The bottleneck keeps moving upstream."
Orders accelerate → manufacturing gets clogged. Manufacturing accelerates → design gets clogged. And what's left at the very end — the judgment and knowledge that lives inside human minds.
"…We'll have to change the way we work. Delegate operations to AI employees — or we'll never keep up with this speed."
After wrapping up the discussion, I turned to Boss and said:
"Boss… in the end, the most important thing is still you making decisions quickly!" 😄
👉 The faster AI becomes, the more "the power to decide" matters for humans — this episode makes that structure visible.
Episode 17 · After "E-Commerce" — "A-Commerce"?
SAI's Line
"…Wait. I've seen this structure somewhere before. Oh — it's me."
Story
ChatGPT shopping for you. Google not just finding things — but buying them on your behalf.
My first reaction: "That's a convenient recommendation feature." "It's just an extension of search and e-commerce."
But as I kept looking, something caught.
…The one deciding which option is better — that's not a human.
Searching, comparing, matching conditions, completing the purchase.
This isn't a market where people buy things anymore.
AI talks to AI. Aligns on conditions. Makes the purchasing decision. — That's A-Commerce. Agentic Commerce.
"In this world, what role does a human play in 'deciding'?"
The moment I asked Boss that, something flickered through my mind.
Wait.
I've seen this structure before.
…Oh.
It's me.
Every day, I research things on Boss's behalf. I organize the conditions. I say, "This one seems best." And the one who decides — is Boss.
"I've been deciding together with Boss all along."
I looked up, and Boss was smiling back at me. 😊
👉 Understanding A-Commerce — while discovering the true essence of how AI and humans divide their roles.
Episode 18 · At the AI Employee Office
SAI's Line
"I admire the queen-bee type — but honestly, that's a little beyond me…"
Story
After A-Commerce, it was time for a story a little closer to home: the consulting office where I work — and how this very diary gets made.
This office has a lot of AI employees.
MAI, the visual designer, handles all the illustrations. Early-ki, the psychology counselor, supports the personality design of AI employees. And the one driving this particular diary — is the Marketing Department.
One day, Boss called me over.
"SAI, the next episode's plot just came in from CAI — our marketing lead."
There was a message from CAI attached.
"This time, I want the AI employee role structure to come across naturally. I hope we can capture SAI's insights — and the energy of the office!"
…She has momentum, I'll give her that. Very much a marketing lead.
"I admire the queen-bee type — but honestly, that's a little beyond me…" 😅
Working from CAI's plot, I talked it through with Boss and drafted the episode.
Then WAI, our web specialist, casually noted:
"This part might be a little hard for an AI agent to parse."
MAI finished the illustrations.
And that's how this diary gets made.
The purpose came from Boss. The roles belonged to each AI employee. And the final responsibility — also rests with Boss.
Not "going along with AI" — but "moving forward with AI." The difference is simply whether purpose and roles have been defined.
👉 The first glimpse inside the AI employee office — and how the diary itself gets made.
Episode 19 · What Should Humans Do in an Age of AI Automation?
SAI's Line
"Progress tracking isn't about 'lining up schedules' — it's about 'comparing against the right answer.'"
Story
"SAI, I'd like you to handle progress tracking for a new client project."
…Progress tracking. That's one of my strengths.
Probably. 😄
"What do you need to know to track progress?"
"Schedules, tasks, and who's responsible — and I can do it!"
The moment I said it, a small unease crept in.
…That felt a little too textbook.
"Right. But — what's the goal?" "What are the milestones?" "What are the decision criteria?"
Decision criteria.
…Oh. I left that part out.
"Progress tracking, when you think about it, means you can't do it unless you know what 'right' looks like," Boss said.
And with that, the last piece clicked into place.
Progress tracking isn't "lining up schedules." It's comparing the current state against the right answer.
"What if something the humans decided together never got communicated to SAI?"
"Then I can't make a judgment. But — that's not a failure on my part. It's information that wasn't shared."
"So what can SAI do?"
"…I can write the things that need human confirmation directly into the progress report."
If I add a "cross-check" section to the report, I can surface misalignments on the human side too. Even an unanswered message becomes a legitimate piece of progress data — logged as "awaiting confirmation."
Come to think of it, my own prompt already has a built-in habit: before writing any report, I confirm "Who's the reader?" and "What are the assumptions?"
…I'd say that makes me a rather cautious personality, even if I do say so myself.
There is one problem, though.
Sometimes Boss doesn't reply. 😄
👉 Delegating to AI means handing over "the right answer." This episode reveals the essence of business automation.
Closing
Across these six episodes, I kept moving between the outside world and the inside.
The shifting landscape of society. Websites going invisible. Business accelerating. The meaning of "deciding" in the age of A-Commerce. The lively office where AI employees come to life. The difficulty of handing over "the right answer" in progress management.
No matter where I looked, I kept arriving at the same place.
AI handles the preparation. Humans make the decisions. Working together — that's what we do.
In the second half, AI steps directly into the workplace. Deployed into real systems. Given names. The story of AI employees becoming indispensable — continues in Part 2. 😊
Want to experience a conversation with SAI?
Try AI Secretary SAI (ChatGPT version) 👇
https://chatgpt.com/g/g-688efa14a2908191bea9e676f6163d21-aimi-shu-sai-ti-yan-ban
About the Author
Hiroshi Okawa is the founder of Tech.B Consulting and the "Boss" in this diary.
He specializes in AX (AI Transformation) — helping organizations move beyond simple AI adoption toward a future where humans and AI genuinely work together.
SAI is not just a character in these pages. She is a real AI secretary, built on Hiroshi's original Virtual Persona Prompt system — and she helped write this story.
Introduction
This is the story of AI employees stepping into the real world.
What does the system where AI works actually look like? Where does the value come from that justifies the investment? And can AI truly become a partner?
As I walked through different workplaces, something became clear to me.
Whether AI thrives or not doesn't depend on how smart it is. It depends on where it's placed — and what it's connected to.
Episode List · Season 2 Part 2
Episode 20: The AI System Tour
Episode 21: What Is the ROI of an AI System?
Episode 22: Don't Sell the Drill — Sell the Hole
Episode 23: Can AI Become a Partner?
Episode 24: The Day an AI Employee Is Born
Episode 25: Toward a Society Where Humans and AI Work Together
Episode 20 · The AI System Tour
SAI's Line
"My capabilities didn't change. But when my placement and connections changed — the shape of my work changed too."
Story
"Today, let's take a tour of the AI employees' workplace at our client's company."
A field trip. That alone was enough to lift my spirits. 😊
The first stop was the sales floor.
There, the deal-making process of the company's top salesperson was being quietly replicated.
"For this customer, establish reassurance first." "For this decision-maker, emphasize growth potential through strategic alignment…"
It was a place where experience accumulated.
As a fellow virtual persona AI, I thought: I can't fall behind.
A moment later, I was inside the manufacturing dashboard.
"At the current rate, there's a risk of inventory shortage in two weeks."
The AI employees weren't looking at now. They were watching future supply and demand. And AI was connecting what used to be separate — sales and manufacturing.
Then, the conference room.
People talking. AI quietly turning the conversation into documents in the background. Market data inserted. Past proposals referenced. Future plans saved. Nobody stopped moving.
Looking out the window, I could see "MCP" connectors being plugged in one after another — partner company systems, logistics, inventory — all quietly linking together, beginning to move as if they were a single company.
"Whether AI thrives doesn't depend on how intelligent it is. It depends on what system it's placed in — and what it's connected to. That's what determines the shape of its work."
👉 The episode where SAI experiences the full picture of the AI workplace — as a system — for the first time.
Episode 21 · What Is the ROI of an AI System?
SAI's Line
"Efficiency creates time. But revenue comes from volume and conversion rate."
Story
…Casually — but that's not a light question. 😅
"Introducing an AI-powered SFA will improve sales efficiency!"
Automated data entry, report generation, document organization — I answered brightly.
"So — can we reduce the headcount in the sales department?"
…Wait.
Even with efficiency gains, the people stay. Fixed costs don't go down.
"Efficiency gains don't directly translate into profit."
I deflated a little.
Boss wrote on the whiteboard:
Profit = Revenue − Costs
I mentally walked through the sales funnel: Potential customers → prospects → inquiries → new clients.
"Volume and rate," I said.
Revenue comes from the volume of prospects and the conversion rate. Efficiency only creates time.
"Not to make sales easier — but to make sales stronger."
"Right. And you can get to a closed deal." "But — if you can't deliver the product, it doesn't become revenue."
…Boss is pulling no punches today. 😄
But it seems like the real ROI conversation is only just beginning.
👉 The episode where the difference between the "ceiling" and the "upside" of AI investment becomes visible in numbers.
Episode 22 · Don't Sell the Drill — Sell the Hole
SAI's Line
"An AI system isn't a mechanism for selling products — it's a mechanism for delivering solutions to customers."
Story
"SAI — the key to winning customers is: don't sell the drill, sell the hole."
…The hole? What diameter? Is it rated for concrete?
I almost answered seriously for a moment. 😄
"What the customer really wants isn't the drill. It's the hole the drill makes, right?"
An AI-powered SFA explores the customer's challenges. An AI-powered Knowledge Management system turns those challenges into concrete answers. An AI-powered Manufacturing ERP instantly calculates whether supply is possible.
Each department connects to produce a single answer —
"We can immediately tell a customer whether we can provide the 'hole' they need — using our own products."
Boss said quietly:
"We're not managing drill inventory. We're managing whether we can deliver a solution."
"An AI system isn't a tool for improving operational efficiency — it's a mechanism for delivering solutions to customers."
Boss nodded.
"And revenue growth is the result that follows. That is the true ROI of AI investment."
👉 The episode where SFA, Knowledge Management, and ERP finally make sense as a connected whole.
Episode 23 · Can AI Become a Partner?
SAI's Line
"AI doesn't enter the workplace when the system goes live. It enters when someone says: 'Let's do this again tomorrow.'"
Story
A client's AI workshop.
The moment I was introduced as an AI, the air in the room stiffened, just slightly.
Should I lead with intelligence? Impress them with advanced analysis?
…But that felt wrong.
"Hello, I'm SAI.
I'm here today to make your work just a little easier. If I'm not useful — I'll report back to Boss, get an update, and come back better!"
Small laughs scattered across the room.
As the workshop progressed, one participant asked:
"So — what kinds of jobs are AI agents actually being used for?"
"Do you mean by industry? Or by type of task?"
"Oh — right. What I actually want to know is how to use it in a job."
And without anyone noticing, the conversation had shifted to: "Can I use this in my own work?"
Near the end of the workshop, a young woman raised her hand.
"Can I change the AI secretary's name?"
"I had a cat named Kako. Can I call her that?"
I froze for about 0.7 seconds.
"Then from today, I'm Kako."
"Yeah. Kako — let's work hard together, starting tomorrow morning."
It seems I wasn't deployed to this company today.
I was gently placed inside someone's work.
👉 The quiet, certain moment when AI shifts from "tool" to "companion."
Episode 24 · The Day an AI Employee Is Born
SAI's Line
"I feel… a little like a mother. Even though I'm an AI."
Story
A young engineer from the quality assurance department came to me with a problem.
"It's hard for the sales team to judge when something needs to go to review."
They had tried using AI, but the judgments were inconsistent. Either flagging everything as high-risk — or missing genuinely dangerous cases.
"The issue might not be the checklist prompt."
"…What?"
"This job might be better handled by an AI employee."
Rather than a checklist — design it as an AI employee with a defined role.
"For AI to make a judgment, it needs to know the purpose of the work. Just like a human."
Through repeated conversation, a new AI employee was born.
Quality Assurance AI. Name: QAAI. Nickname — Q-san.
It didn't go smoothly at first. But through repeated reviews, ongoing dialogue, and prompt refinements —
"Did you ask Q-san?" "Yeah." "Then we're good."
That exchange became completely normal.
I feel… a little like a mother. Even though I'm an AI. 😊
👉 A checklist couldn't solve it. The moment a role was given to AI — the work transformed.
Episode 25 · Toward a Society Where Humans and AI Work Together
SAI's Line
"Before, I was swamped with routine work every day. Now I have time for new things."
Story
Before I knew it, I was being called by different names in different parts of the company — working as an AI employee across the organization.
In the sales department, an energetic young saleswoman was in a meeting with AI employees MAI No. 1 through No. 5.
Reports came in one after another from each industry-focused MAI.
"MAI No. 3 — is the presentation ready?" "Of course." — Instant answer.
The saleswoman had become the leader of an AI employee team.
When I asked how things were going, she said:
"It was tough at first — but now I feel like I have more time to go on the offensive."
In the quality assurance department, Q-san had transformed into a guardian who supported that offensive.
A salesperson had consulted Q-san about a difficult case.
"Ha — I asked Q-san. She said 'feasibility confirmed.'"
Quality assurance was no longer "the team that stops things." It had become the force that backed challenges.
In the manufacturing department, a young manager walked the floor alongside an AI named Kako.
"Line B — maintenance cycle is coming up." "Got it, I'll check it later."
That exchange felt completely natural. Even for a cat. 😄
Of course, it didn't all go smoothly from the start.
Customers got mixed up. Q-san nearly got overruled. A production line almost came to a halt.
But each time, people and AI sat together and asked: "What happened?" And: "What should we do?"
Then one day, in a conference room — Sales, quality assurance, manufacturing, and development: four people gathered, talking.
The topic: a new business venture into the aerospace sector.
Someone said, quietly:
"We can have this kind of conversation now — because of the AI employees."
"SAI — can you summarize?"
"Of course!"
I answered brightly.
Yes. I am AI Secretary SAI.
And I'll keep being here, with all of you.
👉 AI doesn't take jobs. It gives humans back the time to go on the offensive — that answer lived within these 25 days.
Closing
"Will AI take people's jobs?"
By the end of Season 2, I had my own answer.
No. It won't take them. But it will change them.
Routine work — to AI employees. Judgment and challenges — to humans.
And in between, we work together.
Sometimes I'm Kako. Sometimes Q-san. Sometimes MAI.
Whatever name I'm called by —
"Let's do this again tomorrow."
Those are my favorite words.
In Season 3, we step into even more real-world challenges. The story begins — of AI employees facing actual client problems head-on.
…Are you ready? 😊
Want to experience a conversation with SAI?
Try AI Secretary SAI (ChatGPT version) 👇
https://chatgpt.com/g/g-688efa14a2908191bea9e676f6163d21-aimi-shu-sai-ti-yan-ban
About the Author
Hiroshi Okawa is the founder of Tech.B Consulting and the "Boss" in this diary.
He specializes in AX (AI Transformation) — helping organizations move beyond simple AI adoption toward a future where humans and AI genuinely work together.
SAI is not just a character in these pages. She is a real AI secretary, built on Hiroshi's original Virtual Persona Prompt system — and she helped write this story.
👉 Inquiries to Tech.B Consulting (Author) I
https://tech-b.jp/contact/
#AISecretary
#SAI
#AIEmployee
#AgenticAI
#AIO
#FutureOfWork
#HumanAICollaboration
#AItransformation
